1 #ifndef _LINUX_SCHED_H 2 #define _LINUX_SCHED_H 3 4 /* 5 * cloning flags: 6 */ 7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */ 8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */ 9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */ 10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */ 11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */ 12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */ 13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */ 14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */ 15 #define CLONE_THREAD 0x00010000 /* Same thread group? */ 16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */ 17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */ 18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */ 19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */ 20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */ 21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */ 22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */ 23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */ 24 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */ 25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */ 26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */ 27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */ 28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */ 29 #define CLONE_NEWNET 0x40000000 /* New network namespace */ 30 #define CLONE_IO 0x80000000 /* Clone io context */ 31 32 /* 33 * Scheduling policies 34 */ 35 #define SCHED_NORMAL 0 36 #define SCHED_FIFO 1 37 #define SCHED_RR 2 38 #define SCHED_BATCH 3 39 /* SCHED_ISO: reserved but not implemented yet */ 40 #define SCHED_IDLE 5 41 42 #ifdef __KERNEL__ 43 44 struct sched_param { 45 int sched_priority; 46 }; 47 48 #include <asm/param.h> /* for HZ */ 49 50 #include <linux/capability.h> 51 #include <linux/threads.h> 52 #include <linux/kernel.h> 53 #include <linux/types.h> 54 #include <linux/timex.h> 55 #include <linux/jiffies.h> 56 #include <linux/rbtree.h> 57 #include <linux/thread_info.h> 58 #include <linux/cpumask.h> 59 #include <linux/errno.h> 60 #include <linux/nodemask.h> 61 #include <linux/mm_types.h> 62 63 #include <asm/system.h> 64 #include <asm/page.h> 65 #include <asm/ptrace.h> 66 #include <asm/cputime.h> 67 68 #include <linux/smp.h> 69 #include <linux/sem.h> 70 #include <linux/signal.h> 71 #include <linux/path.h> 72 #include <linux/compiler.h> 73 #include <linux/completion.h> 74 #include <linux/pid.h> 75 #include <linux/percpu.h> 76 #include <linux/topology.h> 77 #include <linux/proportions.h> 78 #include <linux/seccomp.h> 79 #include <linux/rcupdate.h> 80 #include <linux/rculist.h> 81 #include <linux/rtmutex.h> 82 83 #include <linux/time.h> 84 #include <linux/param.h> 85 #include <linux/resource.h> 86 #include <linux/timer.h> 87 #include <linux/hrtimer.h> 88 #include <linux/task_io_accounting.h> 89 #include <linux/kobject.h> 90 #include <linux/latencytop.h> 91 #include <linux/cred.h> 92 93 #include <asm/processor.h> 94 95 struct mem_cgroup; 96 struct exec_domain; 97 struct futex_pi_state; 98 struct robust_list_head; 99 struct bio; 100 struct fs_struct; 101 struct bts_context; 102 struct perf_counter_context; 103 104 /* 105 * List of flags we want to share for kernel threads, 106 * if only because they are not used by them anyway. 107 */ 108 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 109 110 /* 111 * These are the constant used to fake the fixed-point load-average 112 * counting. Some notes: 113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 114 * a load-average precision of 10 bits integer + 11 bits fractional 115 * - if you want to count load-averages more often, you need more 116 * precision, or rounding will get you. With 2-second counting freq, 117 * the EXP_n values would be 1981, 2034 and 2043 if still using only 118 * 11 bit fractions. 119 */ 120 extern unsigned long avenrun[]; /* Load averages */ 121 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift); 122 123 #define FSHIFT 11 /* nr of bits of precision */ 124 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 125 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 126 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 127 #define EXP_5 2014 /* 1/exp(5sec/5min) */ 128 #define EXP_15 2037 /* 1/exp(5sec/15min) */ 129 130 #define CALC_LOAD(load,exp,n) \ 131 load *= exp; \ 132 load += n*(FIXED_1-exp); \ 133 load >>= FSHIFT; 134 135 extern unsigned long total_forks; 136 extern int nr_threads; 137 DECLARE_PER_CPU(unsigned long, process_counts); 138 extern int nr_processes(void); 139 extern unsigned long nr_running(void); 140 extern unsigned long nr_uninterruptible(void); 141 extern unsigned long nr_iowait(void); 142 extern void calc_global_load(void); 143 extern u64 cpu_nr_migrations(int cpu); 144 145 extern unsigned long get_parent_ip(unsigned long addr); 146 147 struct seq_file; 148 struct cfs_rq; 149 struct task_group; 150 #ifdef CONFIG_SCHED_DEBUG 151 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 152 extern void proc_sched_set_task(struct task_struct *p); 153 extern void 154 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 155 #else 156 static inline void 157 proc_sched_show_task(struct task_struct *p, struct seq_file *m) 158 { 159 } 160 static inline void proc_sched_set_task(struct task_struct *p) 161 { 162 } 163 static inline void 164 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 165 { 166 } 167 #endif 168 169 extern unsigned long long time_sync_thresh; 170 171 /* 172 * Task state bitmask. NOTE! These bits are also 173 * encoded in fs/proc/array.c: get_task_state(). 174 * 175 * We have two separate sets of flags: task->state 176 * is about runnability, while task->exit_state are 177 * about the task exiting. Confusing, but this way 178 * modifying one set can't modify the other one by 179 * mistake. 180 */ 181 #define TASK_RUNNING 0 182 #define TASK_INTERRUPTIBLE 1 183 #define TASK_UNINTERRUPTIBLE 2 184 #define __TASK_STOPPED 4 185 #define __TASK_TRACED 8 186 /* in tsk->exit_state */ 187 #define EXIT_ZOMBIE 16 188 #define EXIT_DEAD 32 189 /* in tsk->state again */ 190 #define TASK_DEAD 64 191 #define TASK_WAKEKILL 128 192 193 /* Convenience macros for the sake of set_task_state */ 194 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 195 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 196 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 197 198 /* Convenience macros for the sake of wake_up */ 199 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 200 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 201 202 /* get_task_state() */ 203 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 204 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 205 __TASK_TRACED) 206 207 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 208 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 209 #define task_is_stopped_or_traced(task) \ 210 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 211 #define task_contributes_to_load(task) \ 212 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ 213 (task->flags & PF_FROZEN) == 0) 214 215 #define __set_task_state(tsk, state_value) \ 216 do { (tsk)->state = (state_value); } while (0) 217 #define set_task_state(tsk, state_value) \ 218 set_mb((tsk)->state, (state_value)) 219 220 /* 221 * set_current_state() includes a barrier so that the write of current->state 222 * is correctly serialised wrt the caller's subsequent test of whether to 223 * actually sleep: 224 * 225 * set_current_state(TASK_UNINTERRUPTIBLE); 226 * if (do_i_need_to_sleep()) 227 * schedule(); 228 * 229 * If the caller does not need such serialisation then use __set_current_state() 230 */ 231 #define __set_current_state(state_value) \ 232 do { current->state = (state_value); } while (0) 233 #define set_current_state(state_value) \ 234 set_mb(current->state, (state_value)) 235 236 /* Task command name length */ 237 #define TASK_COMM_LEN 16 238 239 #include <linux/spinlock.h> 240 241 /* 242 * This serializes "schedule()" and also protects 243 * the run-queue from deletions/modifications (but 244 * _adding_ to the beginning of the run-queue has 245 * a separate lock). 246 */ 247 extern rwlock_t tasklist_lock; 248 extern spinlock_t mmlist_lock; 249 250 struct task_struct; 251 252 extern void sched_init(void); 253 extern void sched_init_smp(void); 254 extern asmlinkage void schedule_tail(struct task_struct *prev); 255 extern void init_idle(struct task_struct *idle, int cpu); 256 extern void init_idle_bootup_task(struct task_struct *idle); 257 258 extern int runqueue_is_locked(void); 259 extern void task_rq_unlock_wait(struct task_struct *p); 260 261 extern cpumask_var_t nohz_cpu_mask; 262 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 263 extern int select_nohz_load_balancer(int cpu); 264 extern int get_nohz_load_balancer(void); 265 #else 266 static inline int select_nohz_load_balancer(int cpu) 267 { 268 return 0; 269 } 270 #endif 271 272 /* 273 * Only dump TASK_* tasks. (0 for all tasks) 274 */ 275 extern void show_state_filter(unsigned long state_filter); 276 277 static inline void show_state(void) 278 { 279 show_state_filter(0); 280 } 281 282 extern void show_regs(struct pt_regs *); 283 284 /* 285 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 286 * task), SP is the stack pointer of the first frame that should be shown in the back 287 * trace (or NULL if the entire call-chain of the task should be shown). 288 */ 289 extern void show_stack(struct task_struct *task, unsigned long *sp); 290 291 void io_schedule(void); 292 long io_schedule_timeout(long timeout); 293 294 extern void cpu_init (void); 295 extern void trap_init(void); 296 extern void update_process_times(int user); 297 extern void scheduler_tick(void); 298 299 extern void sched_show_task(struct task_struct *p); 300 301 #ifdef CONFIG_DETECT_SOFTLOCKUP 302 extern void softlockup_tick(void); 303 extern void touch_softlockup_watchdog(void); 304 extern void touch_all_softlockup_watchdogs(void); 305 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write, 306 struct file *filp, void __user *buffer, 307 size_t *lenp, loff_t *ppos); 308 extern unsigned int softlockup_panic; 309 extern int softlockup_thresh; 310 #else 311 static inline void softlockup_tick(void) 312 { 313 } 314 static inline void touch_softlockup_watchdog(void) 315 { 316 } 317 static inline void touch_all_softlockup_watchdogs(void) 318 { 319 } 320 #endif 321 322 #ifdef CONFIG_DETECT_HUNG_TASK 323 extern unsigned int sysctl_hung_task_panic; 324 extern unsigned long sysctl_hung_task_check_count; 325 extern unsigned long sysctl_hung_task_timeout_secs; 326 extern unsigned long sysctl_hung_task_warnings; 327 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write, 328 struct file *filp, void __user *buffer, 329 size_t *lenp, loff_t *ppos); 330 #endif 331 332 /* Attach to any functions which should be ignored in wchan output. */ 333 #define __sched __attribute__((__section__(".sched.text"))) 334 335 /* Linker adds these: start and end of __sched functions */ 336 extern char __sched_text_start[], __sched_text_end[]; 337 338 /* Is this address in the __sched functions? */ 339 extern int in_sched_functions(unsigned long addr); 340 341 #define MAX_SCHEDULE_TIMEOUT LONG_MAX 342 extern signed long schedule_timeout(signed long timeout); 343 extern signed long schedule_timeout_interruptible(signed long timeout); 344 extern signed long schedule_timeout_killable(signed long timeout); 345 extern signed long schedule_timeout_uninterruptible(signed long timeout); 346 asmlinkage void __schedule(void); 347 asmlinkage void schedule(void); 348 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner); 349 350 struct nsproxy; 351 struct user_namespace; 352 353 /* Maximum number of active map areas.. This is a random (large) number */ 354 #define DEFAULT_MAX_MAP_COUNT 65536 355 356 extern int sysctl_max_map_count; 357 358 #include <linux/aio.h> 359 360 extern unsigned long 361 arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 362 unsigned long, unsigned long); 363 extern unsigned long 364 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 365 unsigned long len, unsigned long pgoff, 366 unsigned long flags); 367 extern void arch_unmap_area(struct mm_struct *, unsigned long); 368 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 369 370 #if USE_SPLIT_PTLOCKS 371 /* 372 * The mm counters are not protected by its page_table_lock, 373 * so must be incremented atomically. 374 */ 375 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value) 376 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member)) 377 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member) 378 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member) 379 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member) 380 381 #else /* !USE_SPLIT_PTLOCKS */ 382 /* 383 * The mm counters are protected by its page_table_lock, 384 * so can be incremented directly. 385 */ 386 #define set_mm_counter(mm, member, value) (mm)->_##member = (value) 387 #define get_mm_counter(mm, member) ((mm)->_##member) 388 #define add_mm_counter(mm, member, value) (mm)->_##member += (value) 389 #define inc_mm_counter(mm, member) (mm)->_##member++ 390 #define dec_mm_counter(mm, member) (mm)->_##member-- 391 392 #endif /* !USE_SPLIT_PTLOCKS */ 393 394 #define get_mm_rss(mm) \ 395 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss)) 396 #define update_hiwater_rss(mm) do { \ 397 unsigned long _rss = get_mm_rss(mm); \ 398 if ((mm)->hiwater_rss < _rss) \ 399 (mm)->hiwater_rss = _rss; \ 400 } while (0) 401 #define update_hiwater_vm(mm) do { \ 402 if ((mm)->hiwater_vm < (mm)->total_vm) \ 403 (mm)->hiwater_vm = (mm)->total_vm; \ 404 } while (0) 405 406 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) 407 { 408 return max(mm->hiwater_rss, get_mm_rss(mm)); 409 } 410 411 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) 412 { 413 return max(mm->hiwater_vm, mm->total_vm); 414 } 415 416 extern void set_dumpable(struct mm_struct *mm, int value); 417 extern int get_dumpable(struct mm_struct *mm); 418 419 /* mm flags */ 420 /* dumpable bits */ 421 #define MMF_DUMPABLE 0 /* core dump is permitted */ 422 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 423 #define MMF_DUMPABLE_BITS 2 424 425 /* coredump filter bits */ 426 #define MMF_DUMP_ANON_PRIVATE 2 427 #define MMF_DUMP_ANON_SHARED 3 428 #define MMF_DUMP_MAPPED_PRIVATE 4 429 #define MMF_DUMP_MAPPED_SHARED 5 430 #define MMF_DUMP_ELF_HEADERS 6 431 #define MMF_DUMP_HUGETLB_PRIVATE 7 432 #define MMF_DUMP_HUGETLB_SHARED 8 433 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 434 #define MMF_DUMP_FILTER_BITS 7 435 #define MMF_DUMP_FILTER_MASK \ 436 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 437 #define MMF_DUMP_FILTER_DEFAULT \ 438 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ 439 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) 440 441 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS 442 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) 443 #else 444 # define MMF_DUMP_MASK_DEFAULT_ELF 0 445 #endif 446 447 struct sighand_struct { 448 atomic_t count; 449 struct k_sigaction action[_NSIG]; 450 spinlock_t siglock; 451 wait_queue_head_t signalfd_wqh; 452 }; 453 454 struct pacct_struct { 455 int ac_flag; 456 long ac_exitcode; 457 unsigned long ac_mem; 458 cputime_t ac_utime, ac_stime; 459 unsigned long ac_minflt, ac_majflt; 460 }; 461 462 /** 463 * struct task_cputime - collected CPU time counts 464 * @utime: time spent in user mode, in &cputime_t units 465 * @stime: time spent in kernel mode, in &cputime_t units 466 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds 467 * 468 * This structure groups together three kinds of CPU time that are 469 * tracked for threads and thread groups. Most things considering 470 * CPU time want to group these counts together and treat all three 471 * of them in parallel. 472 */ 473 struct task_cputime { 474 cputime_t utime; 475 cputime_t stime; 476 unsigned long long sum_exec_runtime; 477 }; 478 /* Alternate field names when used to cache expirations. */ 479 #define prof_exp stime 480 #define virt_exp utime 481 #define sched_exp sum_exec_runtime 482 483 #define INIT_CPUTIME \ 484 (struct task_cputime) { \ 485 .utime = cputime_zero, \ 486 .stime = cputime_zero, \ 487 .sum_exec_runtime = 0, \ 488 } 489 490 /** 491 * struct thread_group_cputimer - thread group interval timer counts 492 * @cputime: thread group interval timers. 493 * @running: non-zero when there are timers running and 494 * @cputime receives updates. 495 * @lock: lock for fields in this struct. 496 * 497 * This structure contains the version of task_cputime, above, that is 498 * used for thread group CPU timer calculations. 499 */ 500 struct thread_group_cputimer { 501 struct task_cputime cputime; 502 int running; 503 spinlock_t lock; 504 }; 505 506 /* 507 * NOTE! "signal_struct" does not have it's own 508 * locking, because a shared signal_struct always 509 * implies a shared sighand_struct, so locking 510 * sighand_struct is always a proper superset of 511 * the locking of signal_struct. 512 */ 513 struct signal_struct { 514 atomic_t count; 515 atomic_t live; 516 517 wait_queue_head_t wait_chldexit; /* for wait4() */ 518 519 /* current thread group signal load-balancing target: */ 520 struct task_struct *curr_target; 521 522 /* shared signal handling: */ 523 struct sigpending shared_pending; 524 525 /* thread group exit support */ 526 int group_exit_code; 527 /* overloaded: 528 * - notify group_exit_task when ->count is equal to notify_count 529 * - everyone except group_exit_task is stopped during signal delivery 530 * of fatal signals, group_exit_task processes the signal. 531 */ 532 int notify_count; 533 struct task_struct *group_exit_task; 534 535 /* thread group stop support, overloads group_exit_code too */ 536 int group_stop_count; 537 unsigned int flags; /* see SIGNAL_* flags below */ 538 539 /* POSIX.1b Interval Timers */ 540 struct list_head posix_timers; 541 542 /* ITIMER_REAL timer for the process */ 543 struct hrtimer real_timer; 544 struct pid *leader_pid; 545 ktime_t it_real_incr; 546 547 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */ 548 cputime_t it_prof_expires, it_virt_expires; 549 cputime_t it_prof_incr, it_virt_incr; 550 551 /* 552 * Thread group totals for process CPU timers. 553 * See thread_group_cputimer(), et al, for details. 554 */ 555 struct thread_group_cputimer cputimer; 556 557 /* Earliest-expiration cache. */ 558 struct task_cputime cputime_expires; 559 560 struct list_head cpu_timers[3]; 561 562 struct pid *tty_old_pgrp; 563 564 /* boolean value for session group leader */ 565 int leader; 566 567 struct tty_struct *tty; /* NULL if no tty */ 568 569 /* 570 * Cumulative resource counters for dead threads in the group, 571 * and for reaped dead child processes forked by this group. 572 * Live threads maintain their own counters and add to these 573 * in __exit_signal, except for the group leader. 574 */ 575 cputime_t utime, stime, cutime, cstime; 576 cputime_t gtime; 577 cputime_t cgtime; 578 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 579 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 580 unsigned long inblock, oublock, cinblock, coublock; 581 struct task_io_accounting ioac; 582 583 /* 584 * Cumulative ns of schedule CPU time fo dead threads in the 585 * group, not including a zombie group leader, (This only differs 586 * from jiffies_to_ns(utime + stime) if sched_clock uses something 587 * other than jiffies.) 588 */ 589 unsigned long long sum_sched_runtime; 590 591 /* 592 * We don't bother to synchronize most readers of this at all, 593 * because there is no reader checking a limit that actually needs 594 * to get both rlim_cur and rlim_max atomically, and either one 595 * alone is a single word that can safely be read normally. 596 * getrlimit/setrlimit use task_lock(current->group_leader) to 597 * protect this instead of the siglock, because they really 598 * have no need to disable irqs. 599 */ 600 struct rlimit rlim[RLIM_NLIMITS]; 601 602 #ifdef CONFIG_BSD_PROCESS_ACCT 603 struct pacct_struct pacct; /* per-process accounting information */ 604 #endif 605 #ifdef CONFIG_TASKSTATS 606 struct taskstats *stats; 607 #endif 608 #ifdef CONFIG_AUDIT 609 unsigned audit_tty; 610 struct tty_audit_buf *tty_audit_buf; 611 #endif 612 }; 613 614 /* Context switch must be unlocked if interrupts are to be enabled */ 615 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 616 # define __ARCH_WANT_UNLOCKED_CTXSW 617 #endif 618 619 /* 620 * Bits in flags field of signal_struct. 621 */ 622 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 623 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */ 624 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */ 625 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */ 626 /* 627 * Pending notifications to parent. 628 */ 629 #define SIGNAL_CLD_STOPPED 0x00000010 630 #define SIGNAL_CLD_CONTINUED 0x00000020 631 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 632 633 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 634 635 /* If true, all threads except ->group_exit_task have pending SIGKILL */ 636 static inline int signal_group_exit(const struct signal_struct *sig) 637 { 638 return (sig->flags & SIGNAL_GROUP_EXIT) || 639 (sig->group_exit_task != NULL); 640 } 641 642 /* 643 * Some day this will be a full-fledged user tracking system.. 644 */ 645 struct user_struct { 646 atomic_t __count; /* reference count */ 647 atomic_t processes; /* How many processes does this user have? */ 648 atomic_t files; /* How many open files does this user have? */ 649 atomic_t sigpending; /* How many pending signals does this user have? */ 650 #ifdef CONFIG_INOTIFY_USER 651 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 652 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 653 #endif 654 #ifdef CONFIG_EPOLL 655 atomic_t epoll_watches; /* The number of file descriptors currently watched */ 656 #endif 657 #ifdef CONFIG_POSIX_MQUEUE 658 /* protected by mq_lock */ 659 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 660 #endif 661 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 662 663 #ifdef CONFIG_KEYS 664 struct key *uid_keyring; /* UID specific keyring */ 665 struct key *session_keyring; /* UID's default session keyring */ 666 #endif 667 668 /* Hash table maintenance information */ 669 struct hlist_node uidhash_node; 670 uid_t uid; 671 struct user_namespace *user_ns; 672 673 #ifdef CONFIG_USER_SCHED 674 struct task_group *tg; 675 #ifdef CONFIG_SYSFS 676 struct kobject kobj; 677 struct delayed_work work; 678 #endif 679 #endif 680 681 #ifdef CONFIG_PERF_COUNTERS 682 atomic_long_t locked_vm; 683 #endif 684 }; 685 686 extern int uids_sysfs_init(void); 687 688 extern struct user_struct *find_user(uid_t); 689 690 extern struct user_struct root_user; 691 #define INIT_USER (&root_user) 692 693 694 struct backing_dev_info; 695 struct reclaim_state; 696 697 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 698 struct sched_info { 699 /* cumulative counters */ 700 unsigned long pcount; /* # of times run on this cpu */ 701 unsigned long long run_delay; /* time spent waiting on a runqueue */ 702 703 /* timestamps */ 704 unsigned long long last_arrival,/* when we last ran on a cpu */ 705 last_queued; /* when we were last queued to run */ 706 #ifdef CONFIG_SCHEDSTATS 707 /* BKL stats */ 708 unsigned int bkl_count; 709 #endif 710 }; 711 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 712 713 #ifdef CONFIG_TASK_DELAY_ACCT 714 struct task_delay_info { 715 spinlock_t lock; 716 unsigned int flags; /* Private per-task flags */ 717 718 /* For each stat XXX, add following, aligned appropriately 719 * 720 * struct timespec XXX_start, XXX_end; 721 * u64 XXX_delay; 722 * u32 XXX_count; 723 * 724 * Atomicity of updates to XXX_delay, XXX_count protected by 725 * single lock above (split into XXX_lock if contention is an issue). 726 */ 727 728 /* 729 * XXX_count is incremented on every XXX operation, the delay 730 * associated with the operation is added to XXX_delay. 731 * XXX_delay contains the accumulated delay time in nanoseconds. 732 */ 733 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 734 u64 blkio_delay; /* wait for sync block io completion */ 735 u64 swapin_delay; /* wait for swapin block io completion */ 736 u32 blkio_count; /* total count of the number of sync block */ 737 /* io operations performed */ 738 u32 swapin_count; /* total count of the number of swapin block */ 739 /* io operations performed */ 740 741 struct timespec freepages_start, freepages_end; 742 u64 freepages_delay; /* wait for memory reclaim */ 743 u32 freepages_count; /* total count of memory reclaim */ 744 }; 745 #endif /* CONFIG_TASK_DELAY_ACCT */ 746 747 static inline int sched_info_on(void) 748 { 749 #ifdef CONFIG_SCHEDSTATS 750 return 1; 751 #elif defined(CONFIG_TASK_DELAY_ACCT) 752 extern int delayacct_on; 753 return delayacct_on; 754 #else 755 return 0; 756 #endif 757 } 758 759 enum cpu_idle_type { 760 CPU_IDLE, 761 CPU_NOT_IDLE, 762 CPU_NEWLY_IDLE, 763 CPU_MAX_IDLE_TYPES 764 }; 765 766 /* 767 * sched-domains (multiprocessor balancing) declarations: 768 */ 769 770 /* 771 * Increase resolution of nice-level calculations: 772 */ 773 #define SCHED_LOAD_SHIFT 10 774 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 775 776 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE 777 778 #ifdef CONFIG_SMP 779 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */ 780 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */ 781 #define SD_BALANCE_EXEC 4 /* Balance on exec */ 782 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */ 783 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */ 784 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */ 785 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */ 786 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */ 787 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */ 788 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */ 789 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */ 790 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */ 791 792 enum powersavings_balance_level { 793 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */ 794 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package 795 * first for long running threads 796 */ 797 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle 798 * cpu package for power savings 799 */ 800 MAX_POWERSAVINGS_BALANCE_LEVELS 801 }; 802 803 extern int sched_mc_power_savings, sched_smt_power_savings; 804 805 static inline int sd_balance_for_mc_power(void) 806 { 807 if (sched_smt_power_savings) 808 return SD_POWERSAVINGS_BALANCE; 809 810 return 0; 811 } 812 813 static inline int sd_balance_for_package_power(void) 814 { 815 if (sched_mc_power_savings | sched_smt_power_savings) 816 return SD_POWERSAVINGS_BALANCE; 817 818 return 0; 819 } 820 821 /* 822 * Optimise SD flags for power savings: 823 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings. 824 * Keep default SD flags if sched_{smt,mc}_power_saving=0 825 */ 826 827 static inline int sd_power_saving_flags(void) 828 { 829 if (sched_mc_power_savings | sched_smt_power_savings) 830 return SD_BALANCE_NEWIDLE; 831 832 return 0; 833 } 834 835 struct sched_group { 836 struct sched_group *next; /* Must be a circular list */ 837 838 /* 839 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 840 * single CPU. This is read only (except for setup, hotplug CPU). 841 * Note : Never change cpu_power without recompute its reciprocal 842 */ 843 unsigned int __cpu_power; 844 /* 845 * reciprocal value of cpu_power to avoid expensive divides 846 * (see include/linux/reciprocal_div.h) 847 */ 848 u32 reciprocal_cpu_power; 849 850 /* 851 * The CPUs this group covers. 852 * 853 * NOTE: this field is variable length. (Allocated dynamically 854 * by attaching extra space to the end of the structure, 855 * depending on how many CPUs the kernel has booted up with) 856 * 857 * It is also be embedded into static data structures at build 858 * time. (See 'struct static_sched_group' in kernel/sched.c) 859 */ 860 unsigned long cpumask[0]; 861 }; 862 863 static inline struct cpumask *sched_group_cpus(struct sched_group *sg) 864 { 865 return to_cpumask(sg->cpumask); 866 } 867 868 enum sched_domain_level { 869 SD_LV_NONE = 0, 870 SD_LV_SIBLING, 871 SD_LV_MC, 872 SD_LV_CPU, 873 SD_LV_NODE, 874 SD_LV_ALLNODES, 875 SD_LV_MAX 876 }; 877 878 struct sched_domain_attr { 879 int relax_domain_level; 880 }; 881 882 #define SD_ATTR_INIT (struct sched_domain_attr) { \ 883 .relax_domain_level = -1, \ 884 } 885 886 struct sched_domain { 887 /* These fields must be setup */ 888 struct sched_domain *parent; /* top domain must be null terminated */ 889 struct sched_domain *child; /* bottom domain must be null terminated */ 890 struct sched_group *groups; /* the balancing groups of the domain */ 891 unsigned long min_interval; /* Minimum balance interval ms */ 892 unsigned long max_interval; /* Maximum balance interval ms */ 893 unsigned int busy_factor; /* less balancing by factor if busy */ 894 unsigned int imbalance_pct; /* No balance until over watermark */ 895 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 896 unsigned int busy_idx; 897 unsigned int idle_idx; 898 unsigned int newidle_idx; 899 unsigned int wake_idx; 900 unsigned int forkexec_idx; 901 int flags; /* See SD_* */ 902 enum sched_domain_level level; 903 904 /* Runtime fields. */ 905 unsigned long last_balance; /* init to jiffies. units in jiffies */ 906 unsigned int balance_interval; /* initialise to 1. units in ms. */ 907 unsigned int nr_balance_failed; /* initialise to 0 */ 908 909 u64 last_update; 910 911 #ifdef CONFIG_SCHEDSTATS 912 /* load_balance() stats */ 913 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 914 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 915 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 916 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 917 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 918 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 919 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 920 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 921 922 /* Active load balancing */ 923 unsigned int alb_count; 924 unsigned int alb_failed; 925 unsigned int alb_pushed; 926 927 /* SD_BALANCE_EXEC stats */ 928 unsigned int sbe_count; 929 unsigned int sbe_balanced; 930 unsigned int sbe_pushed; 931 932 /* SD_BALANCE_FORK stats */ 933 unsigned int sbf_count; 934 unsigned int sbf_balanced; 935 unsigned int sbf_pushed; 936 937 /* try_to_wake_up() stats */ 938 unsigned int ttwu_wake_remote; 939 unsigned int ttwu_move_affine; 940 unsigned int ttwu_move_balance; 941 #endif 942 #ifdef CONFIG_SCHED_DEBUG 943 char *name; 944 #endif 945 946 /* 947 * Span of all CPUs in this domain. 948 * 949 * NOTE: this field is variable length. (Allocated dynamically 950 * by attaching extra space to the end of the structure, 951 * depending on how many CPUs the kernel has booted up with) 952 * 953 * It is also be embedded into static data structures at build 954 * time. (See 'struct static_sched_domain' in kernel/sched.c) 955 */ 956 unsigned long span[0]; 957 }; 958 959 static inline struct cpumask *sched_domain_span(struct sched_domain *sd) 960 { 961 return to_cpumask(sd->span); 962 } 963 964 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new, 965 struct sched_domain_attr *dattr_new); 966 967 /* Test a flag in parent sched domain */ 968 static inline int test_sd_parent(struct sched_domain *sd, int flag) 969 { 970 if (sd->parent && (sd->parent->flags & flag)) 971 return 1; 972 973 return 0; 974 } 975 976 #else /* CONFIG_SMP */ 977 978 struct sched_domain_attr; 979 980 static inline void 981 partition_sched_domains(int ndoms_new, struct cpumask *doms_new, 982 struct sched_domain_attr *dattr_new) 983 { 984 } 985 #endif /* !CONFIG_SMP */ 986 987 struct io_context; /* See blkdev.h */ 988 989 990 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 991 extern void prefetch_stack(struct task_struct *t); 992 #else 993 static inline void prefetch_stack(struct task_struct *t) { } 994 #endif 995 996 struct audit_context; /* See audit.c */ 997 struct mempolicy; 998 struct pipe_inode_info; 999 struct uts_namespace; 1000 1001 struct rq; 1002 struct sched_domain; 1003 1004 struct sched_class { 1005 const struct sched_class *next; 1006 1007 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup); 1008 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep); 1009 void (*yield_task) (struct rq *rq); 1010 1011 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync); 1012 1013 struct task_struct * (*pick_next_task) (struct rq *rq); 1014 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 1015 1016 #ifdef CONFIG_SMP 1017 int (*select_task_rq)(struct task_struct *p, int sync); 1018 1019 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu, 1020 struct rq *busiest, unsigned long max_load_move, 1021 struct sched_domain *sd, enum cpu_idle_type idle, 1022 int *all_pinned, int *this_best_prio); 1023 1024 int (*move_one_task) (struct rq *this_rq, int this_cpu, 1025 struct rq *busiest, struct sched_domain *sd, 1026 enum cpu_idle_type idle); 1027 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 1028 int (*needs_post_schedule) (struct rq *this_rq); 1029 void (*post_schedule) (struct rq *this_rq); 1030 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task); 1031 1032 void (*set_cpus_allowed)(struct task_struct *p, 1033 const struct cpumask *newmask); 1034 1035 void (*rq_online)(struct rq *rq); 1036 void (*rq_offline)(struct rq *rq); 1037 #endif 1038 1039 void (*set_curr_task) (struct rq *rq); 1040 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 1041 void (*task_new) (struct rq *rq, struct task_struct *p); 1042 1043 void (*switched_from) (struct rq *this_rq, struct task_struct *task, 1044 int running); 1045 void (*switched_to) (struct rq *this_rq, struct task_struct *task, 1046 int running); 1047 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 1048 int oldprio, int running); 1049 1050 #ifdef CONFIG_FAIR_GROUP_SCHED 1051 void (*moved_group) (struct task_struct *p); 1052 #endif 1053 }; 1054 1055 struct load_weight { 1056 unsigned long weight, inv_weight; 1057 }; 1058 1059 /* 1060 * CFS stats for a schedulable entity (task, task-group etc) 1061 * 1062 * Current field usage histogram: 1063 * 1064 * 4 se->block_start 1065 * 4 se->run_node 1066 * 4 se->sleep_start 1067 * 6 se->load.weight 1068 */ 1069 struct sched_entity { 1070 struct load_weight load; /* for load-balancing */ 1071 struct rb_node run_node; 1072 struct list_head group_node; 1073 unsigned int on_rq; 1074 1075 u64 exec_start; 1076 u64 sum_exec_runtime; 1077 u64 vruntime; 1078 u64 prev_sum_exec_runtime; 1079 1080 u64 last_wakeup; 1081 u64 avg_overlap; 1082 1083 u64 nr_migrations; 1084 1085 u64 start_runtime; 1086 u64 avg_wakeup; 1087 1088 #ifdef CONFIG_SCHEDSTATS 1089 u64 wait_start; 1090 u64 wait_max; 1091 u64 wait_count; 1092 u64 wait_sum; 1093 1094 u64 sleep_start; 1095 u64 sleep_max; 1096 s64 sum_sleep_runtime; 1097 1098 u64 block_start; 1099 u64 block_max; 1100 u64 exec_max; 1101 u64 slice_max; 1102 1103 u64 nr_migrations_cold; 1104 u64 nr_failed_migrations_affine; 1105 u64 nr_failed_migrations_running; 1106 u64 nr_failed_migrations_hot; 1107 u64 nr_forced_migrations; 1108 u64 nr_forced2_migrations; 1109 1110 u64 nr_wakeups; 1111 u64 nr_wakeups_sync; 1112 u64 nr_wakeups_migrate; 1113 u64 nr_wakeups_local; 1114 u64 nr_wakeups_remote; 1115 u64 nr_wakeups_affine; 1116 u64 nr_wakeups_affine_attempts; 1117 u64 nr_wakeups_passive; 1118 u64 nr_wakeups_idle; 1119 #endif 1120 1121 #ifdef CONFIG_FAIR_GROUP_SCHED 1122 struct sched_entity *parent; 1123 /* rq on which this entity is (to be) queued: */ 1124 struct cfs_rq *cfs_rq; 1125 /* rq "owned" by this entity/group: */ 1126 struct cfs_rq *my_q; 1127 #endif 1128 }; 1129 1130 struct sched_rt_entity { 1131 struct list_head run_list; 1132 unsigned long timeout; 1133 unsigned int time_slice; 1134 int nr_cpus_allowed; 1135 1136 struct sched_rt_entity *back; 1137 #ifdef CONFIG_RT_GROUP_SCHED 1138 struct sched_rt_entity *parent; 1139 /* rq on which this entity is (to be) queued: */ 1140 struct rt_rq *rt_rq; 1141 /* rq "owned" by this entity/group: */ 1142 struct rt_rq *my_q; 1143 #endif 1144 }; 1145 1146 struct task_struct { 1147 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1148 void *stack; 1149 atomic_t usage; 1150 unsigned int flags; /* per process flags, defined below */ 1151 unsigned int ptrace; 1152 1153 int lock_depth; /* BKL lock depth */ 1154 1155 #ifdef CONFIG_SMP 1156 #ifdef __ARCH_WANT_UNLOCKED_CTXSW 1157 int oncpu; 1158 #endif 1159 #endif 1160 1161 int prio, static_prio, normal_prio; 1162 unsigned int rt_priority; 1163 const struct sched_class *sched_class; 1164 struct sched_entity se; 1165 struct sched_rt_entity rt; 1166 1167 #ifdef CONFIG_PREEMPT_NOTIFIERS 1168 /* list of struct preempt_notifier: */ 1169 struct hlist_head preempt_notifiers; 1170 #endif 1171 1172 /* 1173 * fpu_counter contains the number of consecutive context switches 1174 * that the FPU is used. If this is over a threshold, the lazy fpu 1175 * saving becomes unlazy to save the trap. This is an unsigned char 1176 * so that after 256 times the counter wraps and the behavior turns 1177 * lazy again; this to deal with bursty apps that only use FPU for 1178 * a short time 1179 */ 1180 unsigned char fpu_counter; 1181 #ifdef CONFIG_BLK_DEV_IO_TRACE 1182 unsigned int btrace_seq; 1183 #endif 1184 1185 unsigned int policy; 1186 cpumask_t cpus_allowed; 1187 1188 #ifdef CONFIG_PREEMPT_RCU 1189 int rcu_read_lock_nesting; 1190 int rcu_flipctr_idx; 1191 #endif /* #ifdef CONFIG_PREEMPT_RCU */ 1192 1193 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1194 struct sched_info sched_info; 1195 #endif 1196 1197 struct list_head tasks; 1198 struct plist_node pushable_tasks; 1199 1200 struct mm_struct *mm, *active_mm; 1201 1202 /* task state */ 1203 struct linux_binfmt *binfmt; 1204 int exit_state; 1205 int exit_code, exit_signal; 1206 int pdeath_signal; /* The signal sent when the parent dies */ 1207 /* ??? */ 1208 unsigned int personality; 1209 unsigned did_exec:1; 1210 unsigned in_execve:1; /* Tell the LSMs that the process is doing an 1211 * execve */ 1212 pid_t pid; 1213 pid_t tgid; 1214 1215 /* Canary value for the -fstack-protector gcc feature */ 1216 unsigned long stack_canary; 1217 1218 /* 1219 * pointers to (original) parent process, youngest child, younger sibling, 1220 * older sibling, respectively. (p->father can be replaced with 1221 * p->real_parent->pid) 1222 */ 1223 struct task_struct *real_parent; /* real parent process */ 1224 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */ 1225 /* 1226 * children/sibling forms the list of my natural children 1227 */ 1228 struct list_head children; /* list of my children */ 1229 struct list_head sibling; /* linkage in my parent's children list */ 1230 struct task_struct *group_leader; /* threadgroup leader */ 1231 1232 /* 1233 * ptraced is the list of tasks this task is using ptrace on. 1234 * This includes both natural children and PTRACE_ATTACH targets. 1235 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1236 */ 1237 struct list_head ptraced; 1238 struct list_head ptrace_entry; 1239 1240 /* 1241 * This is the tracer handle for the ptrace BTS extension. 1242 * This field actually belongs to the ptracer task. 1243 */ 1244 struct bts_context *bts; 1245 1246 /* PID/PID hash table linkage. */ 1247 struct pid_link pids[PIDTYPE_MAX]; 1248 struct list_head thread_group; 1249 1250 struct completion *vfork_done; /* for vfork() */ 1251 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1252 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1253 1254 cputime_t utime, stime, utimescaled, stimescaled; 1255 cputime_t gtime; 1256 cputime_t prev_utime, prev_stime; 1257 unsigned long nvcsw, nivcsw; /* context switch counts */ 1258 struct timespec start_time; /* monotonic time */ 1259 struct timespec real_start_time; /* boot based time */ 1260 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1261 unsigned long min_flt, maj_flt; 1262 1263 struct task_cputime cputime_expires; 1264 struct list_head cpu_timers[3]; 1265 1266 /* process credentials */ 1267 const struct cred *real_cred; /* objective and real subjective task 1268 * credentials (COW) */ 1269 const struct cred *cred; /* effective (overridable) subjective task 1270 * credentials (COW) */ 1271 struct mutex cred_guard_mutex; /* guard against foreign influences on 1272 * credential calculations 1273 * (notably. ptrace) */ 1274 1275 char comm[TASK_COMM_LEN]; /* executable name excluding path 1276 - access with [gs]et_task_comm (which lock 1277 it with task_lock()) 1278 - initialized normally by flush_old_exec */ 1279 /* file system info */ 1280 int link_count, total_link_count; 1281 #ifdef CONFIG_SYSVIPC 1282 /* ipc stuff */ 1283 struct sysv_sem sysvsem; 1284 #endif 1285 #ifdef CONFIG_DETECT_HUNG_TASK 1286 /* hung task detection */ 1287 unsigned long last_switch_count; 1288 #endif 1289 /* CPU-specific state of this task */ 1290 struct thread_struct thread; 1291 /* filesystem information */ 1292 struct fs_struct *fs; 1293 /* open file information */ 1294 struct files_struct *files; 1295 /* namespaces */ 1296 struct nsproxy *nsproxy; 1297 /* signal handlers */ 1298 struct signal_struct *signal; 1299 struct sighand_struct *sighand; 1300 1301 sigset_t blocked, real_blocked; 1302 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1303 struct sigpending pending; 1304 1305 unsigned long sas_ss_sp; 1306 size_t sas_ss_size; 1307 int (*notifier)(void *priv); 1308 void *notifier_data; 1309 sigset_t *notifier_mask; 1310 struct audit_context *audit_context; 1311 #ifdef CONFIG_AUDITSYSCALL 1312 uid_t loginuid; 1313 unsigned int sessionid; 1314 #endif 1315 seccomp_t seccomp; 1316 1317 /* Thread group tracking */ 1318 u32 parent_exec_id; 1319 u32 self_exec_id; 1320 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, 1321 * mempolicy */ 1322 spinlock_t alloc_lock; 1323 1324 #ifdef CONFIG_GENERIC_HARDIRQS 1325 /* IRQ handler threads */ 1326 struct irqaction *irqaction; 1327 #endif 1328 1329 /* Protection of the PI data structures: */ 1330 spinlock_t pi_lock; 1331 1332 #ifdef CONFIG_RT_MUTEXES 1333 /* PI waiters blocked on a rt_mutex held by this task */ 1334 struct plist_head pi_waiters; 1335 /* Deadlock detection and priority inheritance handling */ 1336 struct rt_mutex_waiter *pi_blocked_on; 1337 #endif 1338 1339 #ifdef CONFIG_DEBUG_MUTEXES 1340 /* mutex deadlock detection */ 1341 struct mutex_waiter *blocked_on; 1342 #endif 1343 #ifdef CONFIG_TRACE_IRQFLAGS 1344 unsigned int irq_events; 1345 int hardirqs_enabled; 1346 unsigned long hardirq_enable_ip; 1347 unsigned int hardirq_enable_event; 1348 unsigned long hardirq_disable_ip; 1349 unsigned int hardirq_disable_event; 1350 int softirqs_enabled; 1351 unsigned long softirq_disable_ip; 1352 unsigned int softirq_disable_event; 1353 unsigned long softirq_enable_ip; 1354 unsigned int softirq_enable_event; 1355 int hardirq_context; 1356 int softirq_context; 1357 #endif 1358 #ifdef CONFIG_LOCKDEP 1359 # define MAX_LOCK_DEPTH 48UL 1360 u64 curr_chain_key; 1361 int lockdep_depth; 1362 unsigned int lockdep_recursion; 1363 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1364 gfp_t lockdep_reclaim_gfp; 1365 #endif 1366 1367 /* journalling filesystem info */ 1368 void *journal_info; 1369 1370 /* stacked block device info */ 1371 struct bio *bio_list, **bio_tail; 1372 1373 /* VM state */ 1374 struct reclaim_state *reclaim_state; 1375 1376 struct backing_dev_info *backing_dev_info; 1377 1378 struct io_context *io_context; 1379 1380 unsigned long ptrace_message; 1381 siginfo_t *last_siginfo; /* For ptrace use. */ 1382 struct task_io_accounting ioac; 1383 #if defined(CONFIG_TASK_XACCT) 1384 u64 acct_rss_mem1; /* accumulated rss usage */ 1385 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1386 cputime_t acct_timexpd; /* stime + utime since last update */ 1387 #endif 1388 #ifdef CONFIG_CPUSETS 1389 nodemask_t mems_allowed; /* Protected by alloc_lock */ 1390 int cpuset_mem_spread_rotor; 1391 #endif 1392 #ifdef CONFIG_CGROUPS 1393 /* Control Group info protected by css_set_lock */ 1394 struct css_set *cgroups; 1395 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1396 struct list_head cg_list; 1397 #endif 1398 #ifdef CONFIG_FUTEX 1399 struct robust_list_head __user *robust_list; 1400 #ifdef CONFIG_COMPAT 1401 struct compat_robust_list_head __user *compat_robust_list; 1402 #endif 1403 struct list_head pi_state_list; 1404 struct futex_pi_state *pi_state_cache; 1405 #endif 1406 #ifdef CONFIG_PERF_COUNTERS 1407 struct perf_counter_context *perf_counter_ctxp; 1408 struct mutex perf_counter_mutex; 1409 struct list_head perf_counter_list; 1410 #endif 1411 #ifdef CONFIG_NUMA 1412 struct mempolicy *mempolicy; /* Protected by alloc_lock */ 1413 short il_next; 1414 #endif 1415 atomic_t fs_excl; /* holding fs exclusive resources */ 1416 struct rcu_head rcu; 1417 1418 /* 1419 * cache last used pipe for splice 1420 */ 1421 struct pipe_inode_info *splice_pipe; 1422 #ifdef CONFIG_TASK_DELAY_ACCT 1423 struct task_delay_info *delays; 1424 #endif 1425 #ifdef CONFIG_FAULT_INJECTION 1426 int make_it_fail; 1427 #endif 1428 struct prop_local_single dirties; 1429 #ifdef CONFIG_LATENCYTOP 1430 int latency_record_count; 1431 struct latency_record latency_record[LT_SAVECOUNT]; 1432 #endif 1433 /* 1434 * time slack values; these are used to round up poll() and 1435 * select() etc timeout values. These are in nanoseconds. 1436 */ 1437 unsigned long timer_slack_ns; 1438 unsigned long default_timer_slack_ns; 1439 1440 struct list_head *scm_work_list; 1441 #ifdef CONFIG_FUNCTION_GRAPH_TRACER 1442 /* Index of current stored adress in ret_stack */ 1443 int curr_ret_stack; 1444 /* Stack of return addresses for return function tracing */ 1445 struct ftrace_ret_stack *ret_stack; 1446 /* time stamp for last schedule */ 1447 unsigned long long ftrace_timestamp; 1448 /* 1449 * Number of functions that haven't been traced 1450 * because of depth overrun. 1451 */ 1452 atomic_t trace_overrun; 1453 /* Pause for the tracing */ 1454 atomic_t tracing_graph_pause; 1455 #endif 1456 #ifdef CONFIG_TRACING 1457 /* state flags for use by tracers */ 1458 unsigned long trace; 1459 /* bitmask of trace recursion */ 1460 unsigned long trace_recursion; 1461 #endif /* CONFIG_TRACING */ 1462 }; 1463 1464 /* Future-safe accessor for struct task_struct's cpus_allowed. */ 1465 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed) 1466 1467 /* 1468 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1469 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1470 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1471 * values are inverted: lower p->prio value means higher priority. 1472 * 1473 * The MAX_USER_RT_PRIO value allows the actual maximum 1474 * RT priority to be separate from the value exported to 1475 * user-space. This allows kernel threads to set their 1476 * priority to a value higher than any user task. Note: 1477 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1478 */ 1479 1480 #define MAX_USER_RT_PRIO 100 1481 #define MAX_RT_PRIO MAX_USER_RT_PRIO 1482 1483 #define MAX_PRIO (MAX_RT_PRIO + 40) 1484 #define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1485 1486 static inline int rt_prio(int prio) 1487 { 1488 if (unlikely(prio < MAX_RT_PRIO)) 1489 return 1; 1490 return 0; 1491 } 1492 1493 static inline int rt_task(struct task_struct *p) 1494 { 1495 return rt_prio(p->prio); 1496 } 1497 1498 static inline struct pid *task_pid(struct task_struct *task) 1499 { 1500 return task->pids[PIDTYPE_PID].pid; 1501 } 1502 1503 static inline struct pid *task_tgid(struct task_struct *task) 1504 { 1505 return task->group_leader->pids[PIDTYPE_PID].pid; 1506 } 1507 1508 /* 1509 * Without tasklist or rcu lock it is not safe to dereference 1510 * the result of task_pgrp/task_session even if task == current, 1511 * we can race with another thread doing sys_setsid/sys_setpgid. 1512 */ 1513 static inline struct pid *task_pgrp(struct task_struct *task) 1514 { 1515 return task->group_leader->pids[PIDTYPE_PGID].pid; 1516 } 1517 1518 static inline struct pid *task_session(struct task_struct *task) 1519 { 1520 return task->group_leader->pids[PIDTYPE_SID].pid; 1521 } 1522 1523 struct pid_namespace; 1524 1525 /* 1526 * the helpers to get the task's different pids as they are seen 1527 * from various namespaces 1528 * 1529 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1530 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1531 * current. 1532 * task_xid_nr_ns() : id seen from the ns specified; 1533 * 1534 * set_task_vxid() : assigns a virtual id to a task; 1535 * 1536 * see also pid_nr() etc in include/linux/pid.h 1537 */ 1538 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, 1539 struct pid_namespace *ns); 1540 1541 static inline pid_t task_pid_nr(struct task_struct *tsk) 1542 { 1543 return tsk->pid; 1544 } 1545 1546 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, 1547 struct pid_namespace *ns) 1548 { 1549 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); 1550 } 1551 1552 static inline pid_t task_pid_vnr(struct task_struct *tsk) 1553 { 1554 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); 1555 } 1556 1557 1558 static inline pid_t task_tgid_nr(struct task_struct *tsk) 1559 { 1560 return tsk->tgid; 1561 } 1562 1563 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1564 1565 static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1566 { 1567 return pid_vnr(task_tgid(tsk)); 1568 } 1569 1570 1571 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, 1572 struct pid_namespace *ns) 1573 { 1574 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); 1575 } 1576 1577 static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1578 { 1579 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); 1580 } 1581 1582 1583 static inline pid_t task_session_nr_ns(struct task_struct *tsk, 1584 struct pid_namespace *ns) 1585 { 1586 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); 1587 } 1588 1589 static inline pid_t task_session_vnr(struct task_struct *tsk) 1590 { 1591 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); 1592 } 1593 1594 /* obsolete, do not use */ 1595 static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1596 { 1597 return task_pgrp_nr_ns(tsk, &init_pid_ns); 1598 } 1599 1600 /** 1601 * pid_alive - check that a task structure is not stale 1602 * @p: Task structure to be checked. 1603 * 1604 * Test if a process is not yet dead (at most zombie state) 1605 * If pid_alive fails, then pointers within the task structure 1606 * can be stale and must not be dereferenced. 1607 */ 1608 static inline int pid_alive(struct task_struct *p) 1609 { 1610 return p->pids[PIDTYPE_PID].pid != NULL; 1611 } 1612 1613 /** 1614 * is_global_init - check if a task structure is init 1615 * @tsk: Task structure to be checked. 1616 * 1617 * Check if a task structure is the first user space task the kernel created. 1618 */ 1619 static inline int is_global_init(struct task_struct *tsk) 1620 { 1621 return tsk->pid == 1; 1622 } 1623 1624 /* 1625 * is_container_init: 1626 * check whether in the task is init in its own pid namespace. 1627 */ 1628 extern int is_container_init(struct task_struct *tsk); 1629 1630 extern struct pid *cad_pid; 1631 1632 extern void free_task(struct task_struct *tsk); 1633 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1634 1635 extern void __put_task_struct(struct task_struct *t); 1636 1637 static inline void put_task_struct(struct task_struct *t) 1638 { 1639 if (atomic_dec_and_test(&t->usage)) 1640 __put_task_struct(t); 1641 } 1642 1643 extern cputime_t task_utime(struct task_struct *p); 1644 extern cputime_t task_stime(struct task_struct *p); 1645 extern cputime_t task_gtime(struct task_struct *p); 1646 1647 /* 1648 * Per process flags 1649 */ 1650 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */ 1651 /* Not implemented yet, only for 486*/ 1652 #define PF_STARTING 0x00000002 /* being created */ 1653 #define PF_EXITING 0x00000004 /* getting shut down */ 1654 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1655 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1656 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1657 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1658 #define PF_DUMPCORE 0x00000200 /* dumped core */ 1659 #define PF_SIGNALED 0x00000400 /* killed by a signal */ 1660 #define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1661 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */ 1662 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1663 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1664 #define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1665 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1666 #define PF_KSWAPD 0x00040000 /* I am kswapd */ 1667 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */ 1668 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1669 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1670 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1671 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1672 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1673 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1674 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1675 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1676 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1677 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */ 1678 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */ 1679 1680 /* 1681 * Only the _current_ task can read/write to tsk->flags, but other 1682 * tasks can access tsk->flags in readonly mode for example 1683 * with tsk_used_math (like during threaded core dumping). 1684 * There is however an exception to this rule during ptrace 1685 * or during fork: the ptracer task is allowed to write to the 1686 * child->flags of its traced child (same goes for fork, the parent 1687 * can write to the child->flags), because we're guaranteed the 1688 * child is not running and in turn not changing child->flags 1689 * at the same time the parent does it. 1690 */ 1691 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1692 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1693 #define clear_used_math() clear_stopped_child_used_math(current) 1694 #define set_used_math() set_stopped_child_used_math(current) 1695 #define conditional_stopped_child_used_math(condition, child) \ 1696 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1697 #define conditional_used_math(condition) \ 1698 conditional_stopped_child_used_math(condition, current) 1699 #define copy_to_stopped_child_used_math(child) \ 1700 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1701 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1702 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1703 #define used_math() tsk_used_math(current) 1704 1705 #ifdef CONFIG_SMP 1706 extern int set_cpus_allowed_ptr(struct task_struct *p, 1707 const struct cpumask *new_mask); 1708 #else 1709 static inline int set_cpus_allowed_ptr(struct task_struct *p, 1710 const struct cpumask *new_mask) 1711 { 1712 if (!cpumask_test_cpu(0, new_mask)) 1713 return -EINVAL; 1714 return 0; 1715 } 1716 #endif 1717 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1718 { 1719 return set_cpus_allowed_ptr(p, &new_mask); 1720 } 1721 1722 /* 1723 * Architectures can set this to 1 if they have specified 1724 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, 1725 * but then during bootup it turns out that sched_clock() 1726 * is reliable after all: 1727 */ 1728 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1729 extern int sched_clock_stable; 1730 #endif 1731 1732 extern unsigned long long sched_clock(void); 1733 1734 extern void sched_clock_init(void); 1735 extern u64 sched_clock_cpu(int cpu); 1736 1737 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1738 static inline void sched_clock_tick(void) 1739 { 1740 } 1741 1742 static inline void sched_clock_idle_sleep_event(void) 1743 { 1744 } 1745 1746 static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1747 { 1748 } 1749 #else 1750 extern void sched_clock_tick(void); 1751 extern void sched_clock_idle_sleep_event(void); 1752 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1753 #endif 1754 1755 /* 1756 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu 1757 * clock constructed from sched_clock(): 1758 */ 1759 extern unsigned long long cpu_clock(int cpu); 1760 1761 extern unsigned long long 1762 task_sched_runtime(struct task_struct *task); 1763 extern unsigned long long thread_group_sched_runtime(struct task_struct *task); 1764 1765 /* sched_exec is called by processes performing an exec */ 1766 #ifdef CONFIG_SMP 1767 extern void sched_exec(void); 1768 #else 1769 #define sched_exec() {} 1770 #endif 1771 1772 extern void sched_clock_idle_sleep_event(void); 1773 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1774 1775 #ifdef CONFIG_HOTPLUG_CPU 1776 extern void idle_task_exit(void); 1777 #else 1778 static inline void idle_task_exit(void) {} 1779 #endif 1780 1781 extern void sched_idle_next(void); 1782 1783 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 1784 extern void wake_up_idle_cpu(int cpu); 1785 #else 1786 static inline void wake_up_idle_cpu(int cpu) { } 1787 #endif 1788 1789 extern unsigned int sysctl_sched_latency; 1790 extern unsigned int sysctl_sched_min_granularity; 1791 extern unsigned int sysctl_sched_wakeup_granularity; 1792 extern unsigned int sysctl_sched_shares_ratelimit; 1793 extern unsigned int sysctl_sched_shares_thresh; 1794 #ifdef CONFIG_SCHED_DEBUG 1795 extern unsigned int sysctl_sched_child_runs_first; 1796 extern unsigned int sysctl_sched_features; 1797 extern unsigned int sysctl_sched_migration_cost; 1798 extern unsigned int sysctl_sched_nr_migrate; 1799 extern unsigned int sysctl_timer_migration; 1800 1801 int sched_nr_latency_handler(struct ctl_table *table, int write, 1802 struct file *file, void __user *buffer, size_t *length, 1803 loff_t *ppos); 1804 #endif 1805 #ifdef CONFIG_SCHED_DEBUG 1806 static inline unsigned int get_sysctl_timer_migration(void) 1807 { 1808 return sysctl_timer_migration; 1809 } 1810 #else 1811 static inline unsigned int get_sysctl_timer_migration(void) 1812 { 1813 return 1; 1814 } 1815 #endif 1816 extern unsigned int sysctl_sched_rt_period; 1817 extern int sysctl_sched_rt_runtime; 1818 1819 int sched_rt_handler(struct ctl_table *table, int write, 1820 struct file *filp, void __user *buffer, size_t *lenp, 1821 loff_t *ppos); 1822 1823 extern unsigned int sysctl_sched_compat_yield; 1824 1825 #ifdef CONFIG_RT_MUTEXES 1826 extern int rt_mutex_getprio(struct task_struct *p); 1827 extern void rt_mutex_setprio(struct task_struct *p, int prio); 1828 extern void rt_mutex_adjust_pi(struct task_struct *p); 1829 #else 1830 static inline int rt_mutex_getprio(struct task_struct *p) 1831 { 1832 return p->normal_prio; 1833 } 1834 # define rt_mutex_adjust_pi(p) do { } while (0) 1835 #endif 1836 1837 extern void set_user_nice(struct task_struct *p, long nice); 1838 extern int task_prio(const struct task_struct *p); 1839 extern int task_nice(const struct task_struct *p); 1840 extern int can_nice(const struct task_struct *p, const int nice); 1841 extern int task_curr(const struct task_struct *p); 1842 extern int idle_cpu(int cpu); 1843 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *); 1844 extern int sched_setscheduler_nocheck(struct task_struct *, int, 1845 struct sched_param *); 1846 extern struct task_struct *idle_task(int cpu); 1847 extern struct task_struct *curr_task(int cpu); 1848 extern void set_curr_task(int cpu, struct task_struct *p); 1849 1850 void yield(void); 1851 1852 /* 1853 * The default (Linux) execution domain. 1854 */ 1855 extern struct exec_domain default_exec_domain; 1856 1857 union thread_union { 1858 struct thread_info thread_info; 1859 unsigned long stack[THREAD_SIZE/sizeof(long)]; 1860 }; 1861 1862 #ifndef __HAVE_ARCH_KSTACK_END 1863 static inline int kstack_end(void *addr) 1864 { 1865 /* Reliable end of stack detection: 1866 * Some APM bios versions misalign the stack 1867 */ 1868 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 1869 } 1870 #endif 1871 1872 extern union thread_union init_thread_union; 1873 extern struct task_struct init_task; 1874 1875 extern struct mm_struct init_mm; 1876 1877 extern struct pid_namespace init_pid_ns; 1878 1879 /* 1880 * find a task by one of its numerical ids 1881 * 1882 * find_task_by_pid_type_ns(): 1883 * it is the most generic call - it finds a task by all id, 1884 * type and namespace specified 1885 * find_task_by_pid_ns(): 1886 * finds a task by its pid in the specified namespace 1887 * find_task_by_vpid(): 1888 * finds a task by its virtual pid 1889 * 1890 * see also find_vpid() etc in include/linux/pid.h 1891 */ 1892 1893 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid, 1894 struct pid_namespace *ns); 1895 1896 extern struct task_struct *find_task_by_vpid(pid_t nr); 1897 extern struct task_struct *find_task_by_pid_ns(pid_t nr, 1898 struct pid_namespace *ns); 1899 1900 extern void __set_special_pids(struct pid *pid); 1901 1902 /* per-UID process charging. */ 1903 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 1904 static inline struct user_struct *get_uid(struct user_struct *u) 1905 { 1906 atomic_inc(&u->__count); 1907 return u; 1908 } 1909 extern void free_uid(struct user_struct *); 1910 extern void release_uids(struct user_namespace *ns); 1911 1912 #include <asm/current.h> 1913 1914 extern void do_timer(unsigned long ticks); 1915 1916 extern int wake_up_state(struct task_struct *tsk, unsigned int state); 1917 extern int wake_up_process(struct task_struct *tsk); 1918 extern void wake_up_new_task(struct task_struct *tsk, 1919 unsigned long clone_flags); 1920 #ifdef CONFIG_SMP 1921 extern void kick_process(struct task_struct *tsk); 1922 #else 1923 static inline void kick_process(struct task_struct *tsk) { } 1924 #endif 1925 extern void sched_fork(struct task_struct *p, int clone_flags); 1926 extern void sched_dead(struct task_struct *p); 1927 1928 extern void proc_caches_init(void); 1929 extern void flush_signals(struct task_struct *); 1930 extern void __flush_signals(struct task_struct *); 1931 extern void ignore_signals(struct task_struct *); 1932 extern void flush_signal_handlers(struct task_struct *, int force_default); 1933 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 1934 1935 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 1936 { 1937 unsigned long flags; 1938 int ret; 1939 1940 spin_lock_irqsave(&tsk->sighand->siglock, flags); 1941 ret = dequeue_signal(tsk, mask, info); 1942 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 1943 1944 return ret; 1945 } 1946 1947 extern void block_all_signals(int (*notifier)(void *priv), void *priv, 1948 sigset_t *mask); 1949 extern void unblock_all_signals(void); 1950 extern void release_task(struct task_struct * p); 1951 extern int send_sig_info(int, struct siginfo *, struct task_struct *); 1952 extern int force_sigsegv(int, struct task_struct *); 1953 extern int force_sig_info(int, struct siginfo *, struct task_struct *); 1954 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 1955 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 1956 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32); 1957 extern int kill_pgrp(struct pid *pid, int sig, int priv); 1958 extern int kill_pid(struct pid *pid, int sig, int priv); 1959 extern int kill_proc_info(int, struct siginfo *, pid_t); 1960 extern int do_notify_parent(struct task_struct *, int); 1961 extern void force_sig(int, struct task_struct *); 1962 extern void force_sig_specific(int, struct task_struct *); 1963 extern int send_sig(int, struct task_struct *, int); 1964 extern void zap_other_threads(struct task_struct *p); 1965 extern struct sigqueue *sigqueue_alloc(void); 1966 extern void sigqueue_free(struct sigqueue *); 1967 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 1968 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 1969 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 1970 1971 static inline int kill_cad_pid(int sig, int priv) 1972 { 1973 return kill_pid(cad_pid, sig, priv); 1974 } 1975 1976 /* These can be the second arg to send_sig_info/send_group_sig_info. */ 1977 #define SEND_SIG_NOINFO ((struct siginfo *) 0) 1978 #define SEND_SIG_PRIV ((struct siginfo *) 1) 1979 #define SEND_SIG_FORCED ((struct siginfo *) 2) 1980 1981 static inline int is_si_special(const struct siginfo *info) 1982 { 1983 return info <= SEND_SIG_FORCED; 1984 } 1985 1986 /* True if we are on the alternate signal stack. */ 1987 1988 static inline int on_sig_stack(unsigned long sp) 1989 { 1990 return (sp - current->sas_ss_sp < current->sas_ss_size); 1991 } 1992 1993 static inline int sas_ss_flags(unsigned long sp) 1994 { 1995 return (current->sas_ss_size == 0 ? SS_DISABLE 1996 : on_sig_stack(sp) ? SS_ONSTACK : 0); 1997 } 1998 1999 /* 2000 * Routines for handling mm_structs 2001 */ 2002 extern struct mm_struct * mm_alloc(void); 2003 2004 /* mmdrop drops the mm and the page tables */ 2005 extern void __mmdrop(struct mm_struct *); 2006 static inline void mmdrop(struct mm_struct * mm) 2007 { 2008 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 2009 __mmdrop(mm); 2010 } 2011 2012 /* mmput gets rid of the mappings and all user-space */ 2013 extern void mmput(struct mm_struct *); 2014 /* Grab a reference to a task's mm, if it is not already going away */ 2015 extern struct mm_struct *get_task_mm(struct task_struct *task); 2016 /* Remove the current tasks stale references to the old mm_struct */ 2017 extern void mm_release(struct task_struct *, struct mm_struct *); 2018 /* Allocate a new mm structure and copy contents from tsk->mm */ 2019 extern struct mm_struct *dup_mm(struct task_struct *tsk); 2020 2021 extern int copy_thread(unsigned long, unsigned long, unsigned long, 2022 struct task_struct *, struct pt_regs *); 2023 extern void flush_thread(void); 2024 extern void exit_thread(void); 2025 2026 extern void exit_files(struct task_struct *); 2027 extern void __cleanup_signal(struct signal_struct *); 2028 extern void __cleanup_sighand(struct sighand_struct *); 2029 2030 extern void exit_itimers(struct signal_struct *); 2031 extern void flush_itimer_signals(void); 2032 2033 extern NORET_TYPE void do_group_exit(int); 2034 2035 extern void daemonize(const char *, ...); 2036 extern int allow_signal(int); 2037 extern int disallow_signal(int); 2038 2039 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *); 2040 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 2041 struct task_struct *fork_idle(int); 2042 2043 extern void set_task_comm(struct task_struct *tsk, char *from); 2044 extern char *get_task_comm(char *to, struct task_struct *tsk); 2045 2046 #ifdef CONFIG_SMP 2047 extern void wait_task_context_switch(struct task_struct *p); 2048 extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 2049 #else 2050 static inline void wait_task_context_switch(struct task_struct *p) {} 2051 static inline unsigned long wait_task_inactive(struct task_struct *p, 2052 long match_state) 2053 { 2054 return 1; 2055 } 2056 #endif 2057 2058 #define next_task(p) \ 2059 list_entry_rcu((p)->tasks.next, struct task_struct, tasks) 2060 2061 #define for_each_process(p) \ 2062 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 2063 2064 extern bool is_single_threaded(struct task_struct *); 2065 2066 /* 2067 * Careful: do_each_thread/while_each_thread is a double loop so 2068 * 'break' will not work as expected - use goto instead. 2069 */ 2070 #define do_each_thread(g, t) \ 2071 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 2072 2073 #define while_each_thread(g, t) \ 2074 while ((t = next_thread(t)) != g) 2075 2076 /* de_thread depends on thread_group_leader not being a pid based check */ 2077 #define thread_group_leader(p) (p == p->group_leader) 2078 2079 /* Do to the insanities of de_thread it is possible for a process 2080 * to have the pid of the thread group leader without actually being 2081 * the thread group leader. For iteration through the pids in proc 2082 * all we care about is that we have a task with the appropriate 2083 * pid, we don't actually care if we have the right task. 2084 */ 2085 static inline int has_group_leader_pid(struct task_struct *p) 2086 { 2087 return p->pid == p->tgid; 2088 } 2089 2090 static inline 2091 int same_thread_group(struct task_struct *p1, struct task_struct *p2) 2092 { 2093 return p1->tgid == p2->tgid; 2094 } 2095 2096 static inline struct task_struct *next_thread(const struct task_struct *p) 2097 { 2098 return list_entry_rcu(p->thread_group.next, 2099 struct task_struct, thread_group); 2100 } 2101 2102 static inline int thread_group_empty(struct task_struct *p) 2103 { 2104 return list_empty(&p->thread_group); 2105 } 2106 2107 #define delay_group_leader(p) \ 2108 (thread_group_leader(p) && !thread_group_empty(p)) 2109 2110 static inline int task_detached(struct task_struct *p) 2111 { 2112 return p->exit_signal == -1; 2113 } 2114 2115 /* 2116 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 2117 * subscriptions and synchronises with wait4(). Also used in procfs. Also 2118 * pins the final release of task.io_context. Also protects ->cpuset and 2119 * ->cgroup.subsys[]. 2120 * 2121 * Nests both inside and outside of read_lock(&tasklist_lock). 2122 * It must not be nested with write_lock_irq(&tasklist_lock), 2123 * neither inside nor outside. 2124 */ 2125 static inline void task_lock(struct task_struct *p) 2126 { 2127 spin_lock(&p->alloc_lock); 2128 } 2129 2130 static inline void task_unlock(struct task_struct *p) 2131 { 2132 spin_unlock(&p->alloc_lock); 2133 } 2134 2135 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 2136 unsigned long *flags); 2137 2138 static inline void unlock_task_sighand(struct task_struct *tsk, 2139 unsigned long *flags) 2140 { 2141 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 2142 } 2143 2144 #ifndef __HAVE_THREAD_FUNCTIONS 2145 2146 #define task_thread_info(task) ((struct thread_info *)(task)->stack) 2147 #define task_stack_page(task) ((task)->stack) 2148 2149 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 2150 { 2151 *task_thread_info(p) = *task_thread_info(org); 2152 task_thread_info(p)->task = p; 2153 } 2154 2155 static inline unsigned long *end_of_stack(struct task_struct *p) 2156 { 2157 return (unsigned long *)(task_thread_info(p) + 1); 2158 } 2159 2160 #endif 2161 2162 static inline int object_is_on_stack(void *obj) 2163 { 2164 void *stack = task_stack_page(current); 2165 2166 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 2167 } 2168 2169 extern void thread_info_cache_init(void); 2170 2171 #ifdef CONFIG_DEBUG_STACK_USAGE 2172 static inline unsigned long stack_not_used(struct task_struct *p) 2173 { 2174 unsigned long *n = end_of_stack(p); 2175 2176 do { /* Skip over canary */ 2177 n++; 2178 } while (!*n); 2179 2180 return (unsigned long)n - (unsigned long)end_of_stack(p); 2181 } 2182 #endif 2183 2184 /* set thread flags in other task's structures 2185 * - see asm/thread_info.h for TIF_xxxx flags available 2186 */ 2187 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 2188 { 2189 set_ti_thread_flag(task_thread_info(tsk), flag); 2190 } 2191 2192 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2193 { 2194 clear_ti_thread_flag(task_thread_info(tsk), flag); 2195 } 2196 2197 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2198 { 2199 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2200 } 2201 2202 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2203 { 2204 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2205 } 2206 2207 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2208 { 2209 return test_ti_thread_flag(task_thread_info(tsk), flag); 2210 } 2211 2212 static inline void set_tsk_need_resched(struct task_struct *tsk) 2213 { 2214 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2215 } 2216 2217 static inline void clear_tsk_need_resched(struct task_struct *tsk) 2218 { 2219 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2220 } 2221 2222 static inline int test_tsk_need_resched(struct task_struct *tsk) 2223 { 2224 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2225 } 2226 2227 static inline int restart_syscall(void) 2228 { 2229 set_tsk_thread_flag(current, TIF_SIGPENDING); 2230 return -ERESTARTNOINTR; 2231 } 2232 2233 static inline int signal_pending(struct task_struct *p) 2234 { 2235 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2236 } 2237 2238 extern int __fatal_signal_pending(struct task_struct *p); 2239 2240 static inline int fatal_signal_pending(struct task_struct *p) 2241 { 2242 return signal_pending(p) && __fatal_signal_pending(p); 2243 } 2244 2245 static inline int signal_pending_state(long state, struct task_struct *p) 2246 { 2247 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2248 return 0; 2249 if (!signal_pending(p)) 2250 return 0; 2251 2252 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2253 } 2254 2255 static inline int need_resched(void) 2256 { 2257 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2258 } 2259 2260 /* 2261 * cond_resched() and cond_resched_lock(): latency reduction via 2262 * explicit rescheduling in places that are safe. The return 2263 * value indicates whether a reschedule was done in fact. 2264 * cond_resched_lock() will drop the spinlock before scheduling, 2265 * cond_resched_softirq() will enable bhs before scheduling. 2266 */ 2267 extern int _cond_resched(void); 2268 #ifdef CONFIG_PREEMPT_BKL 2269 static inline int cond_resched(void) 2270 { 2271 return 0; 2272 } 2273 #else 2274 static inline int cond_resched(void) 2275 { 2276 return _cond_resched(); 2277 } 2278 #endif 2279 extern int cond_resched_lock(spinlock_t * lock); 2280 extern int cond_resched_softirq(void); 2281 static inline int cond_resched_bkl(void) 2282 { 2283 return _cond_resched(); 2284 } 2285 2286 /* 2287 * Does a critical section need to be broken due to another 2288 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2289 * but a general need for low latency) 2290 */ 2291 static inline int spin_needbreak(spinlock_t *lock) 2292 { 2293 #ifdef CONFIG_PREEMPT 2294 return spin_is_contended(lock); 2295 #else 2296 return 0; 2297 #endif 2298 } 2299 2300 /* 2301 * Thread group CPU time accounting. 2302 */ 2303 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); 2304 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times); 2305 2306 static inline void thread_group_cputime_init(struct signal_struct *sig) 2307 { 2308 sig->cputimer.cputime = INIT_CPUTIME; 2309 spin_lock_init(&sig->cputimer.lock); 2310 sig->cputimer.running = 0; 2311 } 2312 2313 static inline void thread_group_cputime_free(struct signal_struct *sig) 2314 { 2315 } 2316 2317 /* 2318 * Reevaluate whether the task has signals pending delivery. 2319 * Wake the task if so. 2320 * This is required every time the blocked sigset_t changes. 2321 * callers must hold sighand->siglock. 2322 */ 2323 extern void recalc_sigpending_and_wake(struct task_struct *t); 2324 extern void recalc_sigpending(void); 2325 2326 extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2327 2328 /* 2329 * Wrappers for p->thread_info->cpu access. No-op on UP. 2330 */ 2331 #ifdef CONFIG_SMP 2332 2333 static inline unsigned int task_cpu(const struct task_struct *p) 2334 { 2335 return task_thread_info(p)->cpu; 2336 } 2337 2338 extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2339 2340 #else 2341 2342 static inline unsigned int task_cpu(const struct task_struct *p) 2343 { 2344 return 0; 2345 } 2346 2347 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2348 { 2349 } 2350 2351 #endif /* CONFIG_SMP */ 2352 2353 extern void arch_pick_mmap_layout(struct mm_struct *mm); 2354 2355 #ifdef CONFIG_TRACING 2356 extern void 2357 __trace_special(void *__tr, void *__data, 2358 unsigned long arg1, unsigned long arg2, unsigned long arg3); 2359 #else 2360 static inline void 2361 __trace_special(void *__tr, void *__data, 2362 unsigned long arg1, unsigned long arg2, unsigned long arg3) 2363 { 2364 } 2365 #endif 2366 2367 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); 2368 extern long sched_getaffinity(pid_t pid, struct cpumask *mask); 2369 2370 extern void normalize_rt_tasks(void); 2371 2372 #ifdef CONFIG_GROUP_SCHED 2373 2374 extern struct task_group init_task_group; 2375 #ifdef CONFIG_USER_SCHED 2376 extern struct task_group root_task_group; 2377 extern void set_tg_uid(struct user_struct *user); 2378 #endif 2379 2380 extern struct task_group *sched_create_group(struct task_group *parent); 2381 extern void sched_destroy_group(struct task_group *tg); 2382 extern void sched_move_task(struct task_struct *tsk); 2383 #ifdef CONFIG_FAIR_GROUP_SCHED 2384 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2385 extern unsigned long sched_group_shares(struct task_group *tg); 2386 #endif 2387 #ifdef CONFIG_RT_GROUP_SCHED 2388 extern int sched_group_set_rt_runtime(struct task_group *tg, 2389 long rt_runtime_us); 2390 extern long sched_group_rt_runtime(struct task_group *tg); 2391 extern int sched_group_set_rt_period(struct task_group *tg, 2392 long rt_period_us); 2393 extern long sched_group_rt_period(struct task_group *tg); 2394 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk); 2395 #endif 2396 #endif 2397 2398 extern int task_can_switch_user(struct user_struct *up, 2399 struct task_struct *tsk); 2400 2401 #ifdef CONFIG_TASK_XACCT 2402 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2403 { 2404 tsk->ioac.rchar += amt; 2405 } 2406 2407 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2408 { 2409 tsk->ioac.wchar += amt; 2410 } 2411 2412 static inline void inc_syscr(struct task_struct *tsk) 2413 { 2414 tsk->ioac.syscr++; 2415 } 2416 2417 static inline void inc_syscw(struct task_struct *tsk) 2418 { 2419 tsk->ioac.syscw++; 2420 } 2421 #else 2422 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2423 { 2424 } 2425 2426 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2427 { 2428 } 2429 2430 static inline void inc_syscr(struct task_struct *tsk) 2431 { 2432 } 2433 2434 static inline void inc_syscw(struct task_struct *tsk) 2435 { 2436 } 2437 #endif 2438 2439 #ifndef TASK_SIZE_OF 2440 #define TASK_SIZE_OF(tsk) TASK_SIZE 2441 #endif 2442 2443 /* 2444 * Call the function if the target task is executing on a CPU right now: 2445 */ 2446 extern void task_oncpu_function_call(struct task_struct *p, 2447 void (*func) (void *info), void *info); 2448 2449 2450 #ifdef CONFIG_MM_OWNER 2451 extern void mm_update_next_owner(struct mm_struct *mm); 2452 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2453 #else 2454 static inline void mm_update_next_owner(struct mm_struct *mm) 2455 { 2456 } 2457 2458 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2459 { 2460 } 2461 #endif /* CONFIG_MM_OWNER */ 2462 2463 #define TASK_STATE_TO_CHAR_STR "RSDTtZX" 2464 2465 #endif /* __KERNEL__ */ 2466 2467 #endif 2468